Working memory is the ability to retain and manipulate information when it is no longer present in the environment, and is a fundamental requirement for numerous cognitive abilities like decision-making and language comprehension. The prefrontal cortex plays a critical role in working memory, and it has been shown that neurons fire selectively to cues that must be remembered when they are removed, providing evidence that working memory may be held in this region of the brain. Recent work has shown that the firing patterns of neurons during the delay period shows complex dynamics that are not accounted for by many classical theories of prefrontal cortex function. I conjecture that working memory is maintained by small, spatially clustered, sub-populations of cells that activate temporarily before activating another sub- population, like a series of dominos, where the sequence reflects what is remembered and the domino symbolizes the sub-population. This conjecture is based upon recent studies and my preliminary results showing that the majority of neurons represent the memory only transiently and that pairs of neurons seem to interact briefly during a trial. I propose two specific aims designed to test this conjecture and provide information towards a more mechanistic understanding of how prefrontal-cortical circuits maintain activity during working memory tasks. The first specific aim is to test the hypothesis that working memory is represented by a sequence of discrete states, where a state is associated with a small neuronal subpopulation that activates transiently during the delay period, and that different memoranda evoke different sequences of states, their preferred stimulus. The second specific aim is to test the hypothesis that similarly tuned neurons in prefrontal cortex or neurons associated with a state show spatial clustering. These specific aims will be addressed through a combination of chronically implanted arrays of tetrodes and in vivo two-photon imaging while monkeys are engaged in a working memory task. These studies will provide a mechanistic insight into who working memory contents are represented and maintained across a population of neurons. This is important because dysfunctions of working memory are associated with schizophrenia and attention-deficit/hyperactivity disorder. By better understanding working memory maintenance, it may open the door for more targeted pharmacology with improved side-effect profiles.